-Determination of gold in ores furnace atomic absorption by fl.ame

Transcription

-Determination of gold in ores furnace atomic absorption by fl.ame
Analyfica Chimicu Arta, 252 (1991) 97-105
Elscvicr Science Publishers B.V., Amsterdam
97
-Determination of gold in ores by fl.ame and graphite
furnace atomic absorption spectrometry using
a vanadium chemical modifier
C. Garcia-Olalla
Depcrtntent
and A.J. Aller *
of Biochemistry and Molecular Biology. University of L.&n. 24071 ikin
(Received
19th February
1991; revised manuscript
received
(Spain)
16th May 1991)
Abstract
The determination
of gold in ores by both flame and graphite furnace atomic absorption
spectrometry
a.t different
analytical
wavelengths
are compared.
Vanadyl
chloride
was shown to be effective
as a chemical
modifier
for
determining
gold in the presence of heavy metals. A sensitivity of 1 pg I-’ and a detection limit of 1 pg 1-l (or 0.1 pg
6 -1 in the original sample) were obtained. The best linear range was between 1 and 120 pg 1-l and a characteristic
mass of 1 pg u:as obtained.
Keywords:
Atomic
absorption
spectrometry;
Electrothermal
As very low contents of gold in ores can be
economically significant, sensitive analytical methods must be available. Owing to the complexity of
the composition of ores, a combination of several
techniques [l] is required for analysis of these
samples, including classical volurtietric and. gravimetric methods, :c-ray diffraction, x-ray fluorescence, differential therma.I analysis, infrared spectrometry, flame atomic absorption spectrometry
(FAAS) and inductively coupled plasma atomic
emission spectroscopy. The fire-assay [2] and the
optical spectrographic
techniques [l] are considered to be insufficiently sensitive, with a detection
limit of 5 pg g-’ for direct current plasma optical
emission spectrometry [3]. Similarly, most FAAS
methods are very useful only in the mg 1-l range
14-71, alt.hough using a preconcen?ration
step by
liquid-liquid
extraction
or fire-assay detection
limits in the range 12-60 fig g-’ are generally
obtained [8-lo]. The best results are shown by
x-ray fluorescence analysis with excitation by total
COO3-2670/91/%03.50
0 1991 - Elsevier
Science
Publishers
atomization;
Geological
materials:
Gold;
Ores; Vanadium
reflectance [ll] and spark-source mass spectrometry [12,13].
Graphite furnace atomic absorption spectrtitietry (GFAAS) has been widely adopted as 6 versatile technique because of its high sensitivity,[14191. Detection limits of 0.3-5 ng g-’ [2!-221 for
GFUS
are similar to those reported For atomic
fluorescence spectrometry
[23] and instrumental
neutron activation analysis [24], ahhough not as
good as those obtained by radiochemical neutron
activation analysis [25].
The determination
of gold in geolc&al m&ials has received considerable attention [4,5,‘7,9+
11,14-18,2O-22,24,25],
but most of the methods
reported describe its detertination
a.fter a separation/ preconcentration
step into isobutyl
methyl ketone (IBMK) or a fire-assay technique.
Therefore, a procedure that eliminates the necessity for liquid-liquid
extraction or fire-assay preconcentration
has significant advantages. Direct
determinations of gold resulting from samples after
B.V. AH rights reserved
C. GARCIA-GLALLA
decompositions
are usually not applicable as a
result of the complex matrix, particularly constituted by the platinum group metals and other
heavy metals. In this respect it is worth mentioning ihat vanadium in perchlo&
acid medium has
been proposed as an effective releasing agent in
the removal of interferences in FAAS [S]. This
work was aimed at the use of this modifier for the
determination of gold in a natural matrix using
both FAAS and GFAAS.
AND
A.J. ALLER
50.01
mg). Tests to compare nitrous oxideacetylene
and air-acetylene
flames were done
using the 50-mm grooved b-umer in all measurements.
EXPERIMENTAL
Reugen ts
Working solutions were prepared by conventional dilution of stock so1utior.s containing 1000
pg ml-’ of gold. inorganic acids and other chemicals were of analytical-reagent
grade. Distilled,
deionized water from a Mini-Q water purification
system was used for the preparation of samples
and standards.
Instrumentation
All AAS measurements were made on a Therm0
Jarrel Ash SH-11 spectrophotometer
equiped with
an CTF-188 controlled-temperature
graphite furtiace atomiser, using pyrolytic graphite-coated cylindrical graphite tubes. The spectrometer
was
provided with a Smith-Hieftje
background-correction system. A Visimax II gold hollow-cathode
lamp and an Epson 118 recorder were used. Samples were injected by means of an automatic
nebulizer (FASTAC)
system and argon served as
the purge gas. Operational parameters are given in
Table 1. Samples were weighed using a Mettler
AE 240 semimicro analytical balance (sensitivity
Procedure
Ore concentrate
samples were ground to an
average particle size of 70 pm and drih;d in air at
105” C for 2 h. About 1.0 g of finely powdered
sample was accurately weighed, decomposed with
hydrofluoric acid-aqu,a regia (1 + 1) and the resulting solution was evaporated to dryness at lOO110 OC. After cooling and the addition of 15 ml of
0.1 M perchloric acid the samp!c solution was
again reduced to almost dryness and redissolved
in 10 ml of 0.1 M perchloric acid. After warming
and stirring to dissolve residue, the soluiion was
washed into a 25-ml calibrated
flask. Vanadyl
chloride was added to this solution at a final
TABLE 1
Instrumental
parameters
_.
Wavelength (nm)
Spectral band width (nm)
Lamp tiurrcnt (mA)
Background correction
242.8, 367.6, 274.8
0.1
5
Smith-Hieftje
FAAS parameters
Aceiylene flow-rate (1 min - ’ )
Nitrous oxide flow-rate (I miti-‘)
Air flow-rate (I min-‘)
GFAAS parameters
FASTAC delay (s)
Injection volume (c(l)
3.0
9.0
9.0
6.0
10
Furnace programme
WY
Temperature ( o C)
Ramp time (s)
Hold time (s)
Purge (position)
150
2
0
1
Ash 1
450
20
0
2
Ash 2
600-900
20
0
1
Atomize 12
2200
0
4
0
Integration
Clean
2400
0
3
-
DETERMINATION
Ol= GOLD
IN ORES
concentration of 0.1% and the mixture was diluted
to volume with distilled, deionized water. Gold
was subsequently
determined
by FAAS
or
GFAAS. The effect of any drift in sensitivity was
avoided by randomizing measurements and using
the mean of three values.
X-ray fluorescence analysis was done as follows. A 50-ml sample volume was placed over t,he
quartz glass target and dried in a desiccator to
form a thin film Gold was determined at measuring times of 200 s through the La! line, using the
Ka line of Co as an internal standard.
good possibilities for use as an agent counteracting interferences in the determination of gold
by FAAS.
From Table 2, it can be seen that the sensitivity
change factor between wavelengths of 242.8 and
267.6 nm is similar for both flame types, which
agrees with the value reported [26] for gold standard solutions. However; for the wavelength of
274.8 nm this factor is dependent on the flame
used.
As FAAS methods are only useful in the mg
1-l range, graphite furnace atomization will be a
suitable alternative for determining very low levels
of gold.
RESULTS AND DISCUSSION
Del’ermination of gold by GFAAS
Matrix composition effects were studied iu the
determination of gold by GFAAS. Thus, the effect
of metals such as Al, Ca, Cr, Fe, Mgj Si and Te,
usually present in these matrix kypes, was investigated by comparing the responses of the goldmetal solution with those of standard gold solutio;ts. Metals were added either as chlorides or
nitrates. For analytical
applications
it may be
concluded that a chemical modifier must be used
in order to overcome these interferences and/or
increase the analytical signal of gold. Chemical,
modification and furnace conditions were initially
investigated
by using aqueous standards
and
spiked ore concentrate samples.
~~etermination of gold by FAAS
Table 2 summarizes the analytical results at
three analytical wavelengths for the determination
of gold in ores by FAAS using vanadium as releasing agent. Gold can be determined with the optimum. sensitivity using the 242%nm
resonance
wavelength with either the air-acetylene
or the
nitrous oxide-acetylene
flame. Although the sensitivity for gold is better with the air-acetylene
flame, the nitrous oxide-acetylene
flame can be
advantageous in complex matrices. The sensitivity
obtained for the determination
of gold in ore
samples in the presence of vanadium is similar to
that obtained for the gold standard solutions
without vanadium [l.O, 2.2 and 120.0 (mg l-I)-’
for wavelengths of 242.8, 267.6 and 274.8 nm,
respectively] using the air-acetylene
flame. Similar
results were obtained with the nitrous oxideacetylene flame. In conclusion, vanadium shows
Effect of vanadium concentrtition
The suitability of vanadium as a chemical modifier has recently been examined [27] for several
analytes. Vanadium was chosen as a potential
chemical modifier because an isomorphous sub-
TABLE 2
Figures of merit for the FAAS method
Pammeter
Sensitivity (mg I-‘)-’
Detection limit (mg 1-l)
Linear range (mg I- ‘)
’ A-A = air-acetylene
2423 nm
267.6 nm
274.8 nm
A-A 8
N-A ’
A-A ’
N-A ’
A-A a
N-A a
1.15
2.0
2-6fi
2.7
8.0
S-120
2.5
6.0
6-120
5.0
75.0
IS-240
118.2
120
120-4ooo
57.6
80
80-2000
flame; N-A = nitrous oxide-acetylene
flame.
100
C. GARCiA-OLALLA
AND dr.J. ALLER
I.
1.6-
b
a
E
2 1.2 z
::
o
01 0.8 .g
a
L
0.G a
1
I
I
0.’
I
,
I
1
‘Jonodyi
concentration
chloride
I
2
concenlrolionl
on the atomization
700
Ashing
%
of (a) gold standard
stitution in the crystal lattice of the vanadium
species by numerous other elements ce..noccur.
The effect of vanadyl chloride concentration on
the absorbance signal of 1 ng of gold is shown in
Fig. 1. If vanadium is added to the gold standard
solution the enhancement is dependent on the
V/Au ratio, but if it is added to the ore concentrate solution the enhancement becomes constki when the concentration of vanadium exceeds
0.1%. However, the vanadium concentration in
gold ore concentrate solution must not exceed this
600
I
1
0.5
Fig. 1. Effect of vanadium
I
I
800
and (b) gold ore solutions
(100 pg 1-l).
value, because above 0.1% (v/v) the absorptiontime peak profile of the gold signal deteriorates as
consequence of the appearance of several peaks
with absorption time.
Effect of ashing temperature
Ashing plots for aqueous standards and ore
samples are shown in Fig. 2, using wanadyl chloride as chemical modifier. The effect of vanadium
on the thermal stabiliiation of gold in ore concentrate solutions is good, because the addition of
WO
ternperoture,*C
Fig. 2. Peak area in absorbance units of gold solutions (100 pg 1-l) as a function of ashing temperature. (A) Au standard solutions
Gthout modifier; (B) Au standard solutions with 0.1% vanadium as modifier; (C) Au ore concentrate solutions without modifier; (D)
Au arc concentrate solutions with 0.1% vanadium as modifier.
DETERMINATION
OF GOLD
IN ORES
1oi
TABLE 3
Characteristic mass and PH/PA
ratio for (A) gold standard
solutions and (B) ore samples, (a) without and (b) with
vanadium chemical modifier, with an ashing temperature of
800°C
Parameter
A-a
A-b
B-a
B-b
Characteristic
mass (pg)
PI-I/PA ratio
5.0
0.982
4.0
0.912
1.1
0.908
0.4
0.573
vanadium improves the maximum ashing temperature obtained for gold standard solutions. The
ashing temperature can be raised to about 800900” C without loss of gold if vanadium is present. Ore concentrate solutions could tolerate higher
ashing temperatures than aqueous gold solutions;
higher temperatures lead to a shorter delay time
(Fig. 3). The sensitivity of .the gold atomic absorption signal is better in the! presence of vanadium
and, therefore, this chemical modifier was chosen
for determining this element in ores.
In addition to ashing temperature,
the peak
height/peak
area (PH/PA)
ratio and the characteristic mass are parameters used to evaluate the
effectiveness
of a chemical modifier. A good
chemical modifier must show low values for both
of these parameters, as vanadium does (Table 3).
1
2
Enhancement of absorbance
Absorbances increased 3.0- and 2.3-fold were
obtained for vanadium using ashing temperatures
of 600 and 800 o C, respectively, when the relative
absorbance reading for gold in the modifier-free
ore solution was set at 1.0. These enhancement
factors increase to LO- and 8.0-foid when compared with the gold absorbance of the modifierfree gold standard solution. The sensitivity for
gold in ore concentrate solutions is much better
than for standard solutions if vanadium is present.
These results suggest that vanadium is a much
more efficient modifier in the. presence of another
element such as Te or Pt. However, this was not
explored further.
Peak profile characteristics
The absorbance-time
profiles for gold standard
solutions without vanadium modifier show a peak
time at 0.9 s (ashing temperature 600 o C) and 0.6 s
(ashing temperature
800” C) (Fig. 3). H.owever,
the absorbance-time
profiles for the gold ore sample alone and spiked with gold standards in the
absence of vanadium result in a double peak at 6.9
and I.4 s (Fig. 4). The presence of vanadium in
the gold ore samples shifts the time of the second
peak to 1.7 s, whereas the intensity of the first
peak which now appears at 1.1 s is strongly increased (Fig. 5). The peak area and height of both
3
lime,
c
S
5
Fig. 3. Peak profile of gold standard solution (120 pg l-‘)
(b) 8OO*C.
without vanadium
as modifier at an ashing temperature of (a) 600°C
and
C. GARCiA-OLALLA
?02
1
2
4
3
Time,
AND A.J. ALLER
5
3
Fig. 4. Peak profiie of gold (25 pg I-‘) ore samples (a) alone and (b) spiked with gold standards (50 pg I-‘) without vanadium as
chemical modifier. Lines (c) and (d) are the gold “signals pIus background” corresponding to profiles (a) and (b), respectively. Ashing
temperature, 600 o C.
vanadium
reacts preferentially
with the analyte
but not with matrix elements accompaying gold.
The profiles presented in Figs. 4 and 5 cannot
be explained by the roll-over effect. When the
phenomenon
of roll-over is present,
the absorbance vs. time curve should show a dip at the
position of the maximum of the conventional AAS
signal [28-301, which is not the case. However, at
very high concentrations
the roll-over effect may
appear (Fig; 6). For very high concentrations
of
gold (2 mg I-‘), the roll-over is observed for both
peaks appearing for the conventional
AAS signal
(Fig. 6A), whereas for lower concentrations
of
geld (1.2 mg I-‘), this eifect is seen for only one
signals are dependent on the gold and vanadium
concentration.
The intensity of the peaks at 1.1
and 1.7 s for the gold ore concentrates
without
vanadium and spiked with gold standards show a
similar increase (Fig. 4), whereas in the presence
of vanadium the intensity of the first peak (at 1.1
s) increases strongly and that of the second peak
(at: 1.7 s) remains nearly constant or even disappears (Figs. 4 and 5).
When gold ore concentrate
solutions without
vanadium
were analysed the background
absorbance was very important (Fig. 4), whereas a
lower background
signal was obtained
using
vanadium as chemical modifier (Fig. 5). Probably
1
2
3
4
S
Time, S
Fig. 5. (a) Peak profile of gold (40 pg 1-l) ore sample with 0.1% vanadium
temperature profile of graphite tube. Ashing temperature, 600DC.
as chemical modifier; (b) gold signal plus background;
(c)
DETERMllriATiON
OF GOLD
IN ORES
103
TABL.E 4
Figures of merii for the proposed GFAAS method
Wavelength (nm)
Sensitivity(1*gl-‘)
-’
Detection limit {pg I-‘)
Linear range (pg 1-r)
242.8
1
1
I-120
267.6
3
4
4-200
274.8
36.7 (mg I-‘)-’
50 mgl-’
50-8&I mg I-’
of these peaks (Fig. 6B). Roll-over is also observed
at a different wavelength (Fig. 6C), but not at
higher ashirtg temperatures (Fig. 6D),
The sensitivity loss due to the pulsed hollowcathode system depends on the amount of gold,
the wavelength and the ashing temperature, but is
usually in the range 15-40% As a result, this
sensitivity loss has few implications for the anaiytical working range.
Figures of merit
Table 4 gives the sensitivities, detection limits
and linear range of the proposed GFAAS method
1.s*
(based on peak-area ~easurements~
obtained at
three analytical wavelengths. The detection Iimit is
taken as the amount equivalent to. Wee times the
standard detiaticn
obtained for ore concentrate
solution oontaining gold at low leiels, expressed
on the basis of the mass of ore used. Twenty
determinations
were made and the detection limit
was calculated as 0.1 pg g-’ when 1 g of sample
was used.
The sensitivity for 1% absorption (0.0044 absorbance unit) was found to be very good for the
wavelength of 242.8 nm. Note that the sensitivity
change factors cbtained for GFAAS differ considerably from those obtained by FAAS.
Recovery studies and appticatiorrs
Recoveries were investigated both by calibration using gold acid standards and by the method
of standard additions on the gold ore concentrate
samples, using vanadium as chemical modifier in
both instances. Ore concentrate
solutions were
A
1.0.
u7 0.5
d
cv‘
ii
Dm
9
::
1.5
1.0
0.5
Time,s
Pig. 6. Peak profile of gold ore sample without (dashed lines} and with (solid lines) background correction according to
Smith-Mieftje and with 0.1% vanadium as chemical modifier. The hollow-cathode lamp current setting is S mA for the low-current
and 2 mA for the high-current pulse. Wavelength: (A, B, D) 224.8 nm; (C) 267.6 nm. Concentration: (A) 2 mg I-.‘; (B. C, D) 1.2 mg
I-‘. Ashing temperature: (A, B, C) 600*&Z; (D) 900°C.
C. GARCiA-OLALLA
104
AND s&J. ALLER
presence of vanadium was poor for some samples.
However, a good recovery was obtained when all
samples were determined using the standard addition method.
As no standard reference materiais containing
gold at th.e pg 1-l Ievel were available foi ore
concentrates, the accuricy of thi: det&r&ration
of
gold in these ore concentrates
was checked by
comparing the GFAAS
results with those obtained by x-ray fluorescence (XRi;) (Ta.ble 5). A
comparison of the precision of the methods shows
that GFAAS was superior to XRF.
spiked with increasing amounts of goid by adding
.volumes of the standard solutiofi (Ii000 pg Au jr’)
td untreated. sampies. These samples were then
treated following the recommended procedure and
compared with the results obtained for gold acid
standards. The slope of the. calibration line (absorbance vs. doncentrtition) of the spiked ore concentrate sample thus obtained, using peak-area
and -height measurements, was essentially similar
to that obtained for the gold aqueous standard.
There are, however, some small differences among
the slopes from other ore concentrate samples of
different origin, showing that the matnx effects
are not completely eliminated by the use of 0.1%
vanadyl chloride solution. As a result, the gold
concentration in ores is determined preferably by
reference to matrix-matched
standards or standard additions calibration graphs if we take into
account that the analysis is verified on real ore
concentrate samples which cover a range of diverse matrices ranging from silicate to sulphiderich~ and iron+ich materials _(Table 5).
Table 5 shows the precrsron of the measurements (based on peak absorbances) obtained for
seven replicate analyses of three ore concentrate
samples treated as described. The mean precision
&tamed ranged from 3.5 to 6.2% (R.S.D.). The
accuracy of the method was determined by measuring the recovery using standard additions of
gold to the ore concentrate samples. The recovery
obtained using a calibration line prepared in the
Conclusion
It has been
demonstrated
that gold can be
determined satisfactorily
in ore concentrate
extracts without a preconcentration
step prior to
measurements,
using vanadium as a chemical
modifier. A comparison of the GFAAS method
with XRF confirms its vahdity. The method described has been Shown to be simple and efficient
and it offers several advantages over other altematives, notably the lack of sample pretreatment and
the consequently
increased
sample
handling
capacity. The elimination of fusion and liquidliquid extraction avoids the use of hazardous reagents and reduces the possibilities
for sample
contamination.
By proper selection of the analytical wavelength and the atomization
method (FAAS or
GFAAS), a wide range of gold concentrations can
TABLE5
Recovery, analytical
precision and accuracy study a
-.Sample type
Sample Au (mg kg-‘)
CLb
SAC
17.2
16.7(15.5)
Sulphide-rich
21.5
25.3(26.1)
Iron-rich
10.9
9.7(10.5)
Silicate
.
Amounr
added
(mgkg-‘)
Amount
CLb
10.0
20.0
40.0
10.0
20.0
40.0
10.0
20.0
40.0
28.1
37.8
57.0
29.2
38.7
51.9
20.1
29.5
48.8
a Wavelength 242.8 nm. The values obtained by XRF-are
’ Rest& obtained using the standard addition method.
found (mg kg-‘)
Average recovery (X)
R.S.D.
SAC
CLb
SA’
(W)
26.2(25.1)
37.1(35.7)
56.0(54.8)
35.1(34.2j
46.1(45.0)
64.3(63.2)
19.3(19.2)
30.3(29.1)
48.9(47.8)
100.9
99.3(99.2)
3.5(6.9)
90.1
99.9(96.0)
4.7(11.5)
95.8
99.5(94.6)
6.2(8.95)
given in parentheses.
b Results
obtained
from the caIibration
line.
tiETERMiNATION
OF GOLD IN ORES
be detknined,
which could also be of great interest in the analysis of precious metal sweeps and
related materials.
REFERENCES
1
2
3
4
5
6
7
8
9
10
11
12
13
14
S. Kallmann, Anal. Chem., 56 (1984) 1020A.
S. Kallmann and C. Maul, Talanta, 30 (1983) 21.
J.C. van Loon, Trends Anal. Chem., 4 (1985) 24.
F.M. Tindall, At. Absorpt. NY+& 4 (1965) 392.
M.A. Hildon and G.R. Sully, Anal. Chim. Acta, 54 (1971)
245.
R.C. Mallet, D.C.G. Pearton, E.J. Ring and T.W: Steele.
Talanta, 19 (1972) 181.
I. Tsukahara and M. Tanaka, Talanta, 27 (1980) 655.
E. Adriaenssens and F. Verbeek, At. Absorpt. NewsI., 13
(1974) 4:.
P. Hannaker and T.L. Highes, J. Geochem. Explor., 10
(1978) 169.
A. Parkes and R. Murray, At. Absorpt. NewsI., 18 (1979)
57.
R. Eller, Ph D Thesis, University of Mainz, 1986.
K.H. Welch and A.M. Ure, Anaf. Proc.. (1980) 8.
B. Fu, A.M. Ure and T.S. West, Anal. Chim. Acta, 152
(1983) 95.
A.E. Hubert and T.T. Chao, Talanta, 32 (1985) 568.
105 ..
15 E. Kontas, f-f. Niskavaara and J. Viriasalo, Geostand.
New& 10 (1986) 169.
16 T. Stnfilov and T. Todorovski, At. Spectrosc., 8 (1987) i2.
17 V.K. Jain, P.M. Lall and J.S. Tiwari, At. Specirosc.. 8
(1987) 77.
18 M.F. Benedetti, A.M. De Kersabiec and J. Boulexue. Geostand. Newsl., 11 (1987) 12.
19 R. Eller, F. Ah, G. Tiilg and H.J. Tobschall, &es&&s’ 2..
Anal. Chem., 334 (1989) 723.
20 R.R. Brooks, J. Holzbecker, D.E. Ryan and H.F. Zhang,
At. Spectrosc., 2 (1981) 151.
21 G.P. Sighinolfi. C. Gorgoni and A.H. Mohamed, Geostand.
New& 8 (1984) 25.
22 K. Kritsotakis and H.J. Tobschall, Fresenius’ Z. Anaf.
Chem., 320 (1985) 15.
23 P.L. Larkins, Anal. Chim. Acta, 173 (1985) 77.
24 EL. Hoffman, A.J. Naldrett and J.C. Van Loon, Anal.
Chim. Acta, 102 (1978) 157.
25 H.W. Stockmann, J. Radioanal. Chcni., 78 $983) jo?.
26 MS. Cresser, C.E. O’Grady and !.I,. Mat-r, Prog. Anal. At.
Spectrosc., 8 (1985) 19.
27 D.L. Tsalev, T.A. Dimitrov and P.B. Mandjukov, J. Anal.
At. Spectrom., 5 (1990) 189.
28 L. de Gttlan and M.T.C. de Loos-Vollebregt, Spectrochim.
Acta, Part B, 39 (1984) 1011.
29 M.T.C. de Loos-Voliebregt and L. de Galan, Prog. Anal.
At. Spectrosc., 8 (1985) 47.
30 M.T.C. de Loos-Vollebregt and L. de Galan. Spectrochim.
Acta, Part B. 41 (1986) 597.